Each week during the football season, ESPN’s Dan Graziano pens a column that reacts to the most popular hot takes of the moment, judging whether they’re overreactions or on-point. It’s a useful device I’ve occasionally borrowed. Well, as readers know, the emergence of AI has produced a bonfire of hot takes about what it all means for education. Three and a half years on from the November 2022 unveiling of ChatGPT, let’s try to sort through a handful of them.

Employers want graduates who are comfortable using AI tools, so it only makes sense to weave AI into schooling as much as possible. Why would you spend a lot of time teaching facts and skills that students won’t ever need in their jobs? If AI can write for them at work, they may as well learn to use AI to write for them at school.

Verdict: OVERREACTION. For starters, it’s good to recall that education is more than workforce preparation. (Schools are also tasked with developing talents, cultivating critical faculties, and preparing students for citizenship.) But even setting that aside, the assumption is dubious. It hearkens back to the time many advocates and educators seized on the introduction of the handheld calculator to insist that teaching computation was now an unnecessary waste of time. That was, of course, nuts. It turns out that students who haven’t developed math automaticity are going to have trouble spotting errors or making sense of mortgage payments or credit card offers, no matter what tools they use. Cal Newport, professor of computer science at Georgetown University, has aptly noted that “Writing is to cognitive health what steps are to physical health.” Students build capacity through cognitive strain. Developing a thesis, building an argument, assembling evidence, anticipating reader response, and editing text is how we build the muscles of analysis and communication.

But you can’t deny that AI is dazzling. It can already outperform humans at a staggering array of tasks, and it’s just getting started. It knows everything, it writes effortlessly, and it’s tireless. There’s no point in teaching students information they’ll just forget anyway when AI already knows it and can produce it for them. Instead, schools should focus on skills like critical thinking.

Verdict: MISSES THE POINT. AI is dazzling. Its capacity for autonomous action really is astonishing. That’s not the issue. The question is how to prepare students to navigate the capabilities of the AI era. To this end, educators should keep in mind that the quality of AI output depends immensely on the skills and insight of the people using it and is most useful when those users have deep, fluid knowledge of the subject at hand. You want AI to help plan a manned mission to Mars? Great. You better know enough about orbital dynamics, mass/thrust ratios, and material strength to ask it the right questions.

Students have studied literature, history, geography, geometry, and chemistry for centuries—before the printing press, through the advent of the steam engine, and into the age of the personal computer—because this is the knowledge that helps them make sense of their world. Advances in technology, even staggering ones, don’t change that. Meanwhile, critical thought is swell, but you can’t think critically about nothing. There’s absolutely nothing about rigorous content that should stop schools from cultivating critical thought. In fact, if you’re teaching great literature, history, or math without sparking deep thinking, you’re doing it wrong.

AI will save teachers a lot of valuable time by automating mundane tasks. AI may have some limitations when it comes to classroom instruction, but it can free teachers from rote chores like grading, communications, report filing, and IEPs, which consume a huge chunk of teacher time. Offloading these tasks will enable them to devote more time and energy to coaching and mentoring.

Verdict: SHOULDN’T BE AN OVERREACTION, BUT . . . This should be a no-brainer. When one considers the sheer number of tasks that teachers tackle, handing a chunk of them to an AI agent has real promise. The possibilities are immense. That said, this claim has been made repeatedly about innovations over the years. I’ve sat through dozens of pitch sessions and interviews with vendors promising their new tools would transform teacher work. (Narrator: They didn’t.) This often left teachers cynical, frustrated that new portals, smartboards, and apps had ultimately been more hassle than help. So, we shall see. Moreover, if teachers do lean on AI to craft parent emails or IEPs, it may raise new concerns—such as parents feeling even more disconnected or teachers having only a vague sense of what’s actually inside a polished-looking IEP.

Instruction by powerful, personalized, agentic AI is transformative. As the pandemic illustrated for so many, the school day contains a lot of dead time. If schools can squeeze learning into a tight two hours, it frees up lots of time for dynamic, engaging projects. And given the challenge of finding enough skilled educators, this is the surest way to offer every student the tutoring and support they need.

Verdict: THE JURY IS OUT. The case is compelling. There’s evidence that tutoring can make a big difference for students, but the costs and practical challenges of procuring enough trained, reliable tutors are brutal. AI offers a path out of that conundrum. And there are some potentially promising models (and a long body of research into intelligent tutoring systems) that make it clear this is more than a fantasy. We may well see schools in which AI-aided instruction speeds up cycles of mastery, providing more opportunities for students to practice, get personalized explanations, and receive real-time feedback.

But similar claims have been made many times for a variety of algorithm-fueled interventions and new school models. There were exciting “blended” models that disappointed. Given that we know digital reading is less cognitively sticky than reading on paper, we should be asking tough questions about the shortcomings of digital learning in general. The proliferation of screens, phones, and social media has harmed youth mental health and corroded social attachments. AI could well be different. But 15 years of academic stagnation, troubling trends in youth mental health, and increasing social isolation inevitably raise questions about the developmental consequences of potentially ramping up the amount of time students spend wearing earbuds, staring at a screen, in a one-on-one with AI. The potential is real, but much caution is in order.

School systems and colleges need future-forward, human‑centered action plans that’ll protect human agency while equipping educators and learners to thrive amidst rapid technological change. We’ve entered the era of artificial intelligence, and the only way to keep up is by embracing AI-fueled transformation.

Verdict: A BUZZY MESS IN THE FACE OF UNCERTAINTY. This sort of incoherent jumble of strung-together buzzwords fills my inbox (and probably fills yours). But it’s less a hot take than a coping mechanism, a desperate attempt to sound relevant and up to speed. Technological change seems to bring out every advocate’s inner consultant. Over the past quarter-century, many thousands of “tech integration plans” have been lovingly crafted (or cut-and-pasted) by consultants and school leaders. The results have underwhelmed. These plans have had astonishingly little effect on implementation, classrooms, or students. What they’ve birthed instead is a lot of schools with one-to-one devices, too much screen time, too many barely used tech tools, too many disengaged students, and abysmal academic achievement. In many ways, ed tech has left schools worse off by fostering distractions, elbowing aside printed books, and shortening attention spans.

If AI delivers on the ambitions of its evangelists, it may well yield a future in which many humans will feel increasingly peripheral. There are grave questions about the role that schools and colleges would need to play in such a world. Answering them calls for far more than buzzy TED Talks and jargon-laden planning documents. I can’t help but suspect that the most promising path may entail recommitting to the buzz-free fundamentals of cultural literacy, academic knowledge, device-free contemplation, and in-person interaction.

Frederick Hess is an executive editor of Education Next and the author of the blog “Old School with Rick Hess.”